On multi-agent based urban rail transport ridership forecast system

Abstract

Ridership forecast is one of the important bases for urban rail transport network planning, design, construction and operation. For the shortcomings of traditional ridership forecast in the stiff human-computer interaction forms, the signal forecast model, the low computational efficiency and the intensive labor, multi-agent-based urban rail transport ridership forecast system was designed. The Man-Machine-Agent accepted the data from the users and allocated the forecast task to the Management-Agent, in the collaboration and coordination to the next Data-Evaluation-Agent, Model-Selection-Agent, Ridership-Forecast-Agent, returned the forecast results to the Man-Machine-Agent and gave the users the proposed forecast advice and guidance by the User-Proposed-Agent. The system have the excellence of playing a variety of forecast models for a variety of different conditions, and can satisfy the randomness, non-linear and non-deterministic case for the strong adaptability, robustness and flexibility.